artcg / BEGAN

Boundary Equibilibrium Generative Adversarial Networks Implementation in Tensorflow
MIT License
208 stars 43 forks source link

Training error #24

Open bunn183 opened 1 year ago

bunn183 commented 1 year ago

When input --start epoch=0,ValueError: 'a' must be greater than 0 unless no samples are taken

artcg commented 1 year ago

Hi, thanks for opening an issue! This repository was written when Tensorflow was in beta, and the API was quite different. You could try downgrading your Tensorflow and it should work :)

Although by now the BEGAN model is a bit outdated. The modern diffusion models are the state of the art so far as I know, so if you want to generate high quality images you might prefer to try a diffusion model.

Here is one example https://github.com/CompVis/stable-diffusion

Here is a user friendly web interface https://github.com/AUTOMATIC1111/stable-diffusion-webui